shardkey

A shard key is a field or set of fields used by a distributed database to partition data across multiple storage nodes called shards. The shard key determines where a particular document or row is stored and guides query routing to the relevant shards.

Sharding typically uses one of several partitioning schemes. Range-based sharding maps continuous ranges of shard key values to shards, which can support efficient range queries but may create hotspots if the data is unevenly distributed. Hash-based sharding applies a hash function to the shard key value to achieve uniform data distribution, often at the expense of efficient range queries. Compound shard keys combine multiple fields to improve data distribution and support a broader set of queries.

A well-chosen shard key helps minimize cross-shard operations, balance load, and enable efficient routing by the query layer. Key considerations include presence of the shard key in most queries, write and read patterns, data cardinality, and the risk of skew or hotspots. The key is typically immutable after data insertion in many systems, and it should enable even distribution across shards.

Operationally, a sharded deployment comprises shards, config servers, and query routers. Data is split into chunks and migrated between shards as needed for balance. The shard key choice influences performance, maintenance, and future reconfiguration. In many systems, changing the shard key after sharding is complex and may require re-sharding or data migration.

Common examples include MongoDB, where a collection is sharded on a field such as userId, either in a standard or hashed form to achieve different distribution and query characteristics.